Machine Translated by Google

Introduction

Inverters are static converters providing continuous alternating current conversion.

To obtain an alternating voltage adjustable in frequency and in effective value from a

direct voltage, it is necessary to cut the input voltage and apply it sometimes in one direction, sometimes in

the other to the receiver by an adequate sequence of control of the semiconductors. It is therefore

possible to produce at the converter output an alternating voltage of zero average value.

This voltage can include one or more alternating slots depending on whether it is a

control by alternating a slot or by pulse width modulation control

(MLI).

The input source can be of the current source type or of the voltage source type.

output of the converter, the amplitude of the voltages or currents is controlled as well as their

frequency. Inverters are widely used in several industrial applications

(Speed variation of alternating current electric motors, induction heating,

uninterruptible power supply Ö).

1. Study Different types of switch control

1.1 Different types of switch control

To cut the voltage applied to the inverter input, you must operate on the

control of the switches that make up the inverter. The latter has a very large

importance, because it is this which determines the type of output voltage.

Among the different types of order, there are:

1. rectangular wave command,

2. variable width slot control

3. control ‡ bearings
4. command ‡ MLI

1.1.1 Rectangular wave command

For the waveform of this command, as schematized by figure (1.1.a), the value
The effective voltage is not adjustable and depends on the input DC voltage.

1
Machine Translated by Google

1.1.2 Variable width slot control

This order is obtained by the association of rectangular waves out of phase with one another
report to each other with a variable angle, díoÿ, a three-state output control (figure
(1.1.b)).The effective value of the output voltage is variable by acting only on the duration of the
niche.

1.1.3 Control ‡ bearings

It is formed by the sum or difference of slots of variable width. The shape

general is closest to the sinusoid. It is also called running control
staircase (figure (1.1.c)).

1.1.4 Pulse width modulation control

The output wave consists of a train of pulses of variable width and spacing.
The interest of this type of control is to reduce the harmonic rate in order to approach the wave of
output ‡ the sinusoid therefore becomes easily filterable (figure (1.1.d))

(has) (b)

1 1

t t

-1 -1
Wave command Order in width slots
Variable

(c) (d)
1

t t

-1

Control ‡ bearings Command ‡ PWM

Figure 1.1

2
Machine Translated by Google

1.2 Performance parameter of three-phase inverter

1.2.1 Output signal quality

The spectrum of a rectangular signal includes a fundamental wave and harmonics

of more or less significant amplitude and we seek to reduce as much as possible, the amplitude of the

low-rank harmonics because:

• Higher harmonics are easy to filter.

• On an inductive load, it is the low order harmonics which generate the currents
the most important.

1.2.2 Total Harmonic Distortion (THD) Factor

This factor gives us a qualitative idea about the similarity of the real waveform with its

fundamental component. It is given as follows by the following expression:

2 2
VL ÿVL 1
THD= (1.1)
VL 1

With: VL: Effective value of the output voltage.

VL1 : Effective value of the first harmonic of the output voltage.

I.3 Disadvantages of slot control operation

The phase voltage for full-wave operation includes harmonics

odd of order n= 2K+1 (K=0, 1, 3Ö.) and amplitude Vn=V1/n. harmonics of order three and

multiple of three disappear in the voltage between phases. A number of disadvantages

are linked to the presence of these harmonics on the voltages which supply the motors:

• additional magnetic losses and joules.

• The motor torque is oscillating at low frequencies. Oscillating torques can excite
mechanical resonances on the rotor shaft. These mechanical disturbances are
troublesome in many applications where low speed torque must be controlled with
precision (robotics, machine tools).
• Harmonics of the supply current pollute the supply network.

3
Machine Translated by Google

We see, then, the need to introduce other waveforms to resolve this problem:
are pulse width modulation techniques.

1.4 Pulse width modulation PWM

1.4.1 The main objectives of an MLI

ÿ Obtain in the electric charge currents whose variation is close to a sinusoid

by controlling the evolution of the cyclic ratios and thanks to a high frequency of
switching of switches with respect to the frequency of output voltages;
ÿ Allow fine control of the amplitude of the fundamental of the output voltages
generally over the widest possible range and for a widely varying output frequency
variable.

1.4.2 MLI by Natural Sampling

We create two signals:

• A sinusoidal signal of variable amplitude and frequency called the “reference” Vm.

• A very high frequency triangular signal called the “carrier” Vp.

These two signals are compared. The result of the comparison is used to control the opening and the
closing of power circuit switches.

You

Figure 1.2 The comparator output slots

4
Machine Translated by Google

VS1

+E/2

-E/2

Figure 1.3 Output voltage

1.4.2.1 Modulation characteristic

The essential parameters of the MLI are:

fp
• The modulation index: mf = fm

Díoÿ Ép: the modulation frequency (carrier);

Em: the frequency of the fundamentals of the inverter output quantities;

• the adjustment coefficient;

r = peak value of the fundamental of the desired charging voltage

Output voltage slot amplitude

Increasing mf rejects the first non-zero harmonics towards the frequencies

High and facilitates filtering. But, mf is limited by the switching times of the switches
of the converter and therefore by the minimum pulse width.

1.4.2.2 Fundamental and voltage drop

If m is large enough (>6), the fundamental of the output voltage is practically equal to ‡:

1E
E1 í= r (1.2)
22

E1 - effective value of the fundamental of the output voltage.

By action on r, we can theoretically increase the amplitude of the fundamental from zero to one

maximum corresponding to the cancellation of certain slots of the output voltage.

5
Machine Translated by Google

Usually, the effective value E1 of the fundamental while maintaining ‡ r its maximum value, is

lower than that obtained with a full-wave control, i.e.:

í
1 4E 2
E eff (full wave)=
ÿ
E (1.3)
2 ÿ 2 ÿ

Pulse width modulation therefore produces additional voltage drop

compared to a full-wave modulation.

When m is even or odd but very large, and taking r=1, in relation (1.2), we obtain

SO :

ÿ '1
E ÿÿÿÿÿÿÿ E ÿ/22 ÿ
ÿÿ

'
ÿÿÿ ÿÿÿÿÿ 0.7854 (1.4)
fullÿ wave ÿ
E eff 2 /E ÿ 4
max

We have a voltage waste of 21.46%.

The fundamental of the output voltage can therefore reach 78.54% of the driving voltage

(E/2). In practice, the voltage drop is increased by the exact calculation of the conduction times

minimum and maximum.

I.4.3 MLI ‡ symmetrical regular sampling

Unlike natural PWM, in this technique, the sampling of voltages from

reference (the sinusoid) is made at periodically spaced instants.
This modulation technique is a version of the previous ones where we seek to make the signal
symmetrical at time Te, in order to minimize harmonics.
The modulating wave is sampled at each positive peak of the carrier. This value is
kept constant for a chopping period Te (triangular signal period) using
of a zero-order blocker (ZOB).

6
Machine Translated by Google

The principle of this modulation is presented in figure I.4

2
1Ö

(has)

0 0.00 0.00 0.00 0.00 0.0 0.01 0.01 0.01 0.01 0.0

Here 1

0.9

0.8

0.7

0.6
(b)
0.5

0.4

0.3

0.2

0.1

0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02

Vs
E/2

(c)

ñE/2

Figure I.4 MLI ‡ Symmetrical sampling

a) Curve 1: triangular carrier Vt(t)

Curve 2: sinusoidal reference of phase i (Vref(t) = Vmsin(ÿt)).

Curve 3: sampled version of the modulating wave Vref(k).

b) Control pulse for phase i.

c) Output voltage for phase i, VS (t).
This procedure generates a level wave which is the approximation of the reference
sinusoidal. The level wave is then compared to the triangular carrier.

7
Machine Translated by Google

The different inverter control techniques

I.4.4 MLI ‡ Asymmetric regular sampling

Distortion due to harmonics can be reduced by sampling twice

the sinusoidal reference per period of the triangular wave.
Thus, while keeping the same frequency for the switching of the switches, we can
double the sampling frequency, and improve the response by decreasing the distortion rate.
To do this, a sample is taken every half-period of the MLI (or the
sampling period, is divided by two: (Te=Th/2) and the instant is calculated independently
of passage to 1 and the moment of return to 0.

Figure 1.5 shows the principle.

32 1

E/2

(has)

t
0

-E/20 0.002 0.004 Ci 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02

1
(b)

Vs
E/2

(c) t

-E/2

Figure 1.5 MLI ‡ Asymmetric sampling.

8
Machine Translated by Google

a) Curve 1: triangular carrier Vt(t).

Curve 2: sinusoidal reference of phase i Vref(t) =Vmsinÿt.

Curve 3: sampled version of the wave modulating Vref(k).

b) control pulse for phase i.

c) output voltage Vs (t).

I.4.5 Suboptimal MLI

In three-phase, harmonics can be reduced without reducing the amplitude of the voltages.

output since harmonics of order 3 or multiple of 3 are eliminated from the output voltages.

We can add a harmonic of rank 3 to the sinusoid of frequency É to form the wave

of reference. The 3rd harmonic will be found in the fictitious voltages Va0, Vb0, Vc0 with respect to the

fictitious midpoint 0 of the input, but it disappears in the simple voltages Van, Vbn, Vcn and in

Vab, Vbc, Vca at the output.

The addition of harmonic 3 allows to increase the maximum amplitude of the fundamental in

the output voltages. 1

2

E/2 1

0.8

0.6

0.4

0.2
has
0

-0.2

-0.4

-0.6

-0.8

-1
0.005 0.01 0.015 0.02
ñE/2 0

Here
1
0.8
b 0.6
0.4

0.2
0
-0.2
-0.4
-0.6 -0.8 -1 0 0.00 0.01 0.01 0.0

Figure 1.6 Suboptimal MLI

a) Curve 1: triangular carrier Vt(t)

Curve 2: reference (Va-V0)w b)
Control pulse for phase i.

9
Machine Translated by Google

The control using a reference voltage of the form:

U
=
(Va-Vo) w (r sin(ÿt)+k sin(3ÿt)) 2 (1.5)

is called suboptimal control.

To find the value of k that allows r to reach its maximum value, we assume that

the reference waveform is always the same when writing:

U
(Va-Vo)w = r(sin(ÿt)+k ísin(3ÿt)) (1.6)
2

The derivative with respect to ‡ t of sin (ÿt)+k ísin(3ÿt) shows that the first maximum is

í
1
product before t = /2ÿ if k >
9

If this condition is met, the time of this maximum is such that:

1ÿ 1 ÿ
ÿ

Cos 2ÿt = ÿ3
(1.7)
4 ÿ 3 k' ÿ ÿ

The maximum value of r corresponds to the minimum of the quantity (sin(ÿt)+k ísin(3ÿt)) at that instant,

2
we thus find: rmax = =1.1547 (1.8)
3

2
1k 1 or k= =0.19245 (1.9)
í= 6 6 3

The theoretical maximum amplitude of the fundamental of the output phase-to-phase voltages increases from

E/2, with sinusoidal control ‡ 1.1547 E/2, with suboptimal control.

The voltage drop goes from 21.46ÿ to 9.31ÿ.

10
Machine Translated by Google

I.4.6 Calculated MLI

I.4.6.1 Principle

This PWM technique consists of calculating the switching times of the switches of

way to take up certain criteria relating to the frequency spectrum of the resulting wave. These

sequence are then memorized and cyclically restored to ensure the control of the

switches.

The criteria usually used are:

- Harmonic elimination of specified order.

- Harmonic elimination in a specified frequency band.

I.4.6.2 Fourier series decomposition of a bipolar PWM signal

Generally, a wave is used which has symmetry with respect to the quarter period

then, deduce the other angles by symmetry. Figure 1.7 shows us after decomposition into

Fourier series of a bipolar PWM signal symmetrical with respect to the quarter period and

antisymmetric with respect to the half-period. The existence of odd order harmonics.

The switching angles ·1, ·2, ·3, ·4 allow us, by controlling them, to eliminate

the three harmonics and control the fundamental.

UC

0 ÿ/2 3ÿ/2 2ÿ

ÿ4
ÿ1 ÿ ÿ ÿ4 ÿ ÿ ÿ1
ÿ2ÿ3 ÿ ÿ ÿ2
ÿ ÿ ÿ3

Figure 1.7 Bipolar PWM signal

The Fourier transform of a periodic alternating signal is given by:

0a ÿ ÿ

ÿÿ2 (nt)ÿBÿ sin ÿ

U(t) ÿ
n 1ÿ
has
n sin ÿ
(nt)
n 1ÿ
n (1.10)

11
Machine Translated by Google

T
With : 2
has
n
ÿ ÿ ÿ

T
COS ÿ t dt
2
(1.11)
T
2
and b ÿ ÿ
ÿ

T
sin t dtÿ
n 2

For a periodic signal with symmetry on the quarter period and antisymmetry on the

half period, we have:

atn0 ÿ
And

4 ÿ
2
b ÿ
ÿ sin ÿntÿ ÿ dt (1.12)
n ÿ 0

If we assume that the signal U(t) has an amplitude E equal to +1 then bn will be equal to:

ÿ ÿ 2 ÿ 3
bn ÿ ÿ ÿ ÿ ÿ1sinÿ ÿ ÿ 1sin ÿÿÿ 1ÿ
ntdt ÿ ÿÿÿ ÿ 1sin ÿÿÿ 2ÿ
ntdt ÿ ÿÿ ÿ ntdt
ÿ ÿ ÿ ÿ ...
0 ÿ ÿ

ÿ k ÿ /2
(1.13)
ÿ ÿ ÿ ÿ1sin ÿ ÿ ntÿ ÿdt ÿ ÿ ÿÿ ÿ ÿ1sin ÿ ntÿ ÿdt ÿÿ
ÿ k ÿ

1 ÿ k

And using the following expression:

ÿ2
ÿ
ÿ1
sin ÿnt
ÿÿ ÿ cosÿn ÿ2ÿÿ
ÿ dt 1/ nÿcosÿn ÿ1ÿ (1.14)

The first of the last terms become:

ÿ 1
ÿ ÿÿ1ÿsinÿ
nt dtÿ nÿ ÿ n ÿ 1/ ÿ1 cosÿ ÿ1ÿÿ (1.15)
0
And

ÿ /2 k k

ÿ kÿ ÿ 1ÿsinÿ
n ÿ t dtÿ n 1/ ÿcosÿn ÿ k ÿÿÿ 1ÿ ÿ ÿ

(1.16)
ÿ

By integrating the other terms of equation (1.13) and substituting equations (1.15) and
(1.16), we will have:

4 k
b
n
ÿ

nÿ
ÿ ÿ 1 cos
1 2ÿ ÿÿ ÿcos (n) 1 k) ÿÿ
(n)ÿ ÿ ÿ ÿ ÿ 1 cos (n )1...ÿ ÿ cos(n (1.17)

4 k
k
b
n
ÿ

nÿ
ÿ ÿÿÿ12
k1
ÿ
ÿ 1ÿcos (n ÿk) ÿ (1.18)

12
Machine Translated by Google

Equation (1.18) contains K equation ‡ K unknowns, the fundamental can be controlled and

(k-1) harmonics can be eliminated.

So we have:

U(t) ÿÿ ÿ 1 42 1cos( ) ÿsin(

k
ÿ

nÿ
ÿi ÿ ÿ ÿntÿ ÿ ntÿ ) ÿ
(1.19)
i 1
n 1
ÿ
ÿ

Knowing that the effective value is given by:

b1 b1 bn
U 1ÿ ,U 3 ÿ
,.....,U n ÿ
(1.20)
2 2 2

From the previous equations we obtain the following system of equations:

22 k
ÿ1 2cos 1 ÿ ÿ ÿ ÿ
U 1ÿ
ÿ
( ) 2 ( ) 2 ( ) 2 ( ) ...ÿ ÿ 2cos( k) ÿ
ÿ ÿ cos 2 ÿ ÿ ÿ ÿ cos 3 cos 4 ÿ

22 k
U 3 ÿ ÿ1 2 cos
ÿ
ÿÿ cos
(3) 21 (3) 2 (3) 2
2 (3) ...ÿ
ÿÿ
cos
ÿ 2cos(3 k) 3ÿ
ÿÿ cos ÿ ÿ4 ÿ

1 ÿ (1.21)
3ÿ
ÿ

22 k
U n
ÿ1 2 cos ÿÿ cos ÿÿ
cos ÿÿ cos ÿ ÿ4 ÿ

1 ÿ
(n) 2 1(n) 2 (n) 2
k) ÿ 3
ÿ ÿ

ÿ
2 (n) ...ÿ ÿ 2cos(n

For full-wave control, the fundamental for a bridge assembly has the effective value:

22
U10 ÿ
E (1.22)
ÿ

By substituting equations (1.21) and (1.22) into the above system and comparing the value

effective of the output obtained with the PWM ‡ that of the full-wave control, then in

representing by P the relationship to the fundamental of the full command, the solution sought

is the one that must verify the following condition: 0<ÿ1<ÿ2<ÿ3<Ö<ÿ/2

13
Machine Translated by Google

For the elimination of (k-1) harmonic, we must solve the following system of equations:

U 1 ÿ 1
ÿÿ
cos ( ÿ1)ÿ (ÿ2)
ÿ ÿcos
ÿÿÿÿ cos ( 3) c ( 4) ... cos( k) ÿ
bone
2U 10 ÿÿ
2

ÿ 1
0 ÿÿ
cos(3
ÿ ÿ ÿ1)
ÿ cos(3 2) cos(3
ÿ ÿ 3)
ÿ ÿ ÿcos(3 4) ... cos(3ÿ k) ÿ
ÿÿ 2 (1.23)

0 ÿ ÿ1 ÿ cos(nÿ1) ÿ cos(nÿ2) ÿ cos(nÿ3) ÿ cos(nÿ4) ÿ...ÿ cos(nÿk) ÿ

The solution of this system is possible with the application of one of the iterative methods such as

than that of NEWTON-RAPHSON.

1.4.6.3 The NEWTON-RAPHSON method

*
ÿ Principle: we are looking for the value X which cancels the function F. the NEWTON method-

*
RAPHSON allows you to approximate the X value by means of the following relationship:

XK-1= XK-F(XK)/ Fí(XK) (1.24)

The choice of initial conditions always remains a delicate task because the system
is nonlinear.

ÿ Application of the method : for a three-phase system, the harmonics of rank three or

multiple of three disappear. we calculate the k angles so that the first (k-1) are cancelled

odd harmonics other than those of rank three or a multiple of three The choice for our

application is to eliminate only the first three (5, 7 and 11).

So our system below will consist of four equations with four variables.

p
2
ÿÿÿ

2U
U 1
10
ÿ 21cos( ÿ) cos(
ÿ1 ÿ ) cos(ÿ ) cos(
2 ÿ ÿ

ÿ ) ÿ 3 4

1 0ÿ ÿcos(5)
ÿ ÿ 1cos(5)
cos(5) cos(5)
ÿ ÿ2 ÿ 2ÿ ÿ 3ÿ ÿ 4
(1.25)
1 0ÿ cos(7)
ÿÿ ÿ 1cos(7)
cos(7) cos(7)
ÿ ÿ2 ÿ 2ÿ ÿ 3ÿ ÿ 4

1 0ÿ cos(11
ÿÿ ÿ 1 ÿ) cos(11 ) ÿ 2ÿ 2 ÿ
) cos(11 ) cos(11 ÿ 3ÿ ÿ 4

14
Machine Translated by Google

After solving the system, we obtain the results presented in the following table:

P=U '1 /U '10 Alpha 1 Alpha 2 Alpha 3 Alpha 3

0.1 20.9584 38.6043 61.1352 79.3324
0.2 21.8448 37.1335 62.3461 78.7498
0.3 22.6192 35.5363 63.6574 78.2834
0.4 23.1949 33.7179 65.1106 77.9797
0.5 23.3546 31.4801 66.7711 77.8979
0.6 22.5153 28.4261 68.7146 78.0327
0.7 19.8603 24.3789 70.9260 78.0840
0.8 16.0218 20.3015 73.5546 78.0898
0.9 11.3507 16.2643 79.8117 81.5269

Table (1.1) Calculation of angles by the Newton Raphson method

1.4.7 Vector modulation inverters

1.4.7.1 Principle
+ TA+ TB+ TC+

U/2
=

HAS B C
0

U/2
=

YOUR- TB- TC-

Vn1 ÿ Vn2 ÿ Vn3 ÿ

N
Figure 1.8

The reference voltages are the desired simple voltages Vni.

This MLI technique follows the following principles:

ÿ the reference signal is sampled at regular intervals T (regular PWM), ÿ For each phase,

production of a pulse of width T centered on the period (PWM

symmetrical) whose average value is equal to the reference voltage at the instant

sampling,

15
Machine Translated by Google

ÿ All switches in the same half-bridge have an identical state in the center and at both
ends of the period (for discontinuous PWM, the state of one of the switches of each
half-bridge remains constant which reduces switching losses but increases the
harmonics).
1.4.7.2 Calculation of switching times
ÌAN

The reference voltages are the desired simple voltages VS = ÌBN referenced by
relative to the neutral point. ÌCN

The voltages generated by a PWM are referenced to the fictitious midpoint O of the
inverter source.

The voltages between phases are given by:

ÌAB = ÌAO - ÌBO , ÌBC = ÌBO - ÌCO and ÌCA = ÌCO - ÌAO (1.26)

If the load is balanced: ÌAN + ÌBN + ÌCN = 0

God: ÌAN = 1/3(ÌAB - ÌCA)

ÌBN = 1/3(ÌBC - ÌAB) (1.27)
ÌCN = 1/3(ÌCA - ÌBC)

SO : Ì n1= ÌAN = 1/3(2ÌAO - ÌBO - ÌCO)

= = ÌBN = 1/3(2ÌBO - ÌCO - ÌAO) Ì n2 = ÌCN (1.28)
Ì n3
1/3(2ÌCO - ÌAO - ÌBO)

And the potential of the neutral point is given by: ÌNO = 1/3(ÌAO + ÌBO + ÌCO)

To simplify the calculations and represent these tensions, let us apply the transformation
three-phase/two-phase respecting power transfer:

The Clarke Transformation

The Clarke transformation consists of substituting for the three real variables Ì n1, Ì n2, Ì n3

their components VS·, VS‚ and VS0 are given by:

VS· 1 -1/2 -1/2 Ì n1

= 2 0 3 /2
VS‚ - /2 Ì n2 3 (1.29)
VS0 3 1/2 1/2 1/ 2 Ì n3

16
Machine Translated by Google

The VS0 component is identically zero since the voltages Ì n1, Ì n2, Ì n3 do not contain

of homopolar component:

VS· 1 -1/2 -1/2 Ì n1

=2 Ì n2 (1.30)
VS‚ 3 0 3 /2 - 3 /2 Ì n3

If we consider the two remaining components VS and VS as the projections on two axes

perpendicular to a vector V, this vector alone is sufficient to characterize the three-phase system.

A combinatorial analysis of all possible states of the switches allows us to calculate the

corresponding VS· voltage vectors :

VS‚

+ + +
YOUR TB TC VA0 VB0 VC0 Vn1 Vn2 Vn3 V· V‚

0 0 0 -U/2 -U/2 -U/2 0 0 0 0 0 V0

-U -U
0 0 1 -U/2 -U/2 U/2 -U/3 -U/3 2U/3 V5
6 2
-UU
0 1 0 -U/2 U/2 -U/2 -U/3 2U/3 -U/3 V3
6 2
2
ÿ

U
0 1 1 -U/2 U/2 EU2 -2U/3 U/3 U/3 3 0 V4

2
1 0 0 U/2 -U/2 -U/2 2U/3 -U/3 -U/3 U 0
3 V1

U -U
1 0 1 U/2 -U/2 U/2 U/3 -2U/3 U/3 V6
6 2
U -U
1 1 0 U/2 U/2 -U/2 U/3 U/3 -2U/3 V2
6 2
1 1 1 U/2 U/2 U/2 0 0 0 0 0
V7

Table (1.2) calculation of tension vectors.

The table shows the eight states that the switches of the three-phase bridge can take to six

switches.

Two of these vectors are identically zero. The other six have the same modulus equal to U 2 / 3 .

The ends of these six vectors define the vertices of a regular hexagon, since two

successive vectors make an angle of /3 between them, or even (figure 1.9).

17
Machine Translated by Google

‚U
2
010 110
ÿ II ÿ
V3 V2
III I VS
ÿ ÿ

011 V4 ÿt V1 100
·
IV VI 2 U

ÿVÿ 3
V5 V6

001 101

Figure 1.9

ÿÿ

We will control the voltage vector VS or Vref we can define it:

In three-phase coordinates:
ÌAN = r U/2 cos ÿt
ÌBN = r U/2 cos (ÿt-2/3) (1.31)
ÌCN = r U/2 cos (ÿt-4/3)
In Cartesian coordinates: ÿ

Vref = Ì· + j Ì‚ (1.32)

With:
3
Ì· = r U/2 cos ÿt
2
(1.33)
3
Ì‚ = r U/2 sin ÿt
2

In polar coordinates:
3 I'll
Vref = r U/2 e (1.34)
2

At any time the vector VS can be expressed as a linear combination of the two
output vectors that are adjacent to it
:ÿ
- When the angle ÿt that VS makes with the axis · is between 0 and /3 (FIG 1.10.a) we
has :

3 3
VS = r sin /3 - ÿt V1 + r sin ÿt V2 (1.35)
2 2

- When the angle ÿt is between /3 and 2/3 (FIG 1.10.b) we have:

3 3
VS = r sin 2/3 - ÿt V2 + r sin ÿt - /3 V3 (1.36)
2 2

18
Machine Translated by Google

And so on:

‚ ‚

V2 V3 V2
VS

VS

ÿt V1 ÿt
· ·
has) b)

Figure 1.10

In general:

3 3
VS = r sin (x+1) /3 - ÿt Vx+1 + r sin ÿt - x /3 Vx+2 (1.37)
2 2

Where x represents the integer part of the quotient of ÿt by /3 and where the indices x+1 and x+2 of the vectors
must be taken modulo 6.

If we set ÿtí= ÿt ñ x /3, the previous relation writes:

3 3
VS = r sin /3 ñ ÿtí Vx+1 + r sinÿtí Vx+2 (1.38)
2 2

Note :
The end of the vector VS remains inside the hexagon defined by the ends V1 and V6 :
rÿ2
3
The previous relationship indicates the durations that must be imposed on the two configurations
corresponding to the two output vectors between which is (VS)n and consequently the
time when it is necessary to take a configuration giving a zero output vector.

We have:

(VS)n = 1 ( Tíi Vi + Tíi+1 Vi+1 ) (1.39)

Tm

Tí0 = Tm ñ (Tíi + Tíi+1) (1.40)

Tm : the modulation period.

VS is sampled at the frequency fm = 1/ Tm

19
Machine Translated by Google

2
j (i -1) /3
Vi = Ue (1.41)
3
SO:
2 You j (i -1) /3 e
Ti ÿ1' j
(i /3) e
(VS)n = U + (1.42)
3 Tm Tm

We deduce:

3
Tíi = r Tm sin (i /3 ñ ÿt)
2
3
Tíi+1 = r Tm sin (ÿt ñ (i-1) /3) (1.43)
2
Tí0 = Tm ñ (Tíi + Tíi+1)

V0 Vimp Vpair V7 V7 Vpair Vimp V0

( 1-k0)T0 Timp Tpair k0T0 k0T0 Tpair Timp (1-k0)T0

Tm/2 Tm/2

- Sequence of application of vectors over a period Tm- (k0 factor ™ [0,1]).

20
Machine Translated by Google

Figure 1.11: Description of the conduction sequences of the switches.

21
Machine Translated by Google

1.4.7.3 Definition of the time sequence of state vectors

The time sequence of the state vectors is done with a possible a priori choice in the

repair of null vectors.

These choices made it possible to respond to the constraints:

On harmonics

In the case of an inductive load, the harmonic vectors of voltage and current are related by:

(n+1)Tm (n+1)Tm
ƒi = 1/L ÿ ƒV dt = 1/L ÿ (Vi ñ (Vs)n) dt (1.44)
nTm nTm

So if we put state vectors Vi close to (Vs)n then we reduce the harmonics of

fluent.

On the switching frequency

It is interesting to reduce the number of switchings of each arm. Among the two vectors

adjacent, there is always:

an “odd” state vector: an “even” state vector:

V1 ÿ (1,0,1) V2 ÿ (1,1,0)
V3 ÿ (0,1,0) V4 ÿ (0,1,1)
V5 ÿ (0,0,1) V6 ÿ (1,0,1)

As there is the choice for the realization of a zero vector:

V0 ÿ (0,0,0) is used before and after V1, V3, V5 (odd state vectors)

V7 ÿ (1,1,1) is used before and after V2, V4, V6 (even state vectors)

In this case, an arm only switches twice during the period Tm, the frequency of
switching is therefore equal to fm. This modulation period is broken down into two half-
symmetrical Tm/2 periods .

Thanks to the symmetry of the signals, this PWM method allows access, if necessary, to a
image of the average value of the output currents over the period Tm by sampling their
values in the middle of this period without resorting to filtering.

22
Machine Translated by Google

The durations of application of the vectors are written for a half-period:

3 Tm
Ti = r sin (i /3 ñ ÿt)
2 2

3 Tm
Ti+1 = r sin (ÿt ñ (i-1) /3) (1.45)
2 2

Tm
T0 = ñ (Ti + Ti+1)
2

The choice of sequences is carried out according to the algorithm described in figure 1.12

If VS‚ > 0 then

If VS· > 0 then
sector 2 time time calculation
If VS‚ > 3 VS· then calculation otherwise from sector 1

Otherwise

sector 2 time time calculation

If VS‚ > - 3 VS· then calculation otherwise from sector 3

End if
Otherwise

If VS· > 0 then

sector 5 time time calculation
If VS, < - 3 VS· then calculation otherwise from sector 6

Otherwise

sector 5 time time calculation

If - VS‚ > - 3 VS· then calculation otherwise from sector 4

End if
End if

Figure 1.12: Vector PWM algorithm.

23
Machine Translated by Google

2 Simulation
2.1 Interpretation of curves
In our study we used MLI techniques to see the signal quality at the terminal of

the inverter, using as a continuous source E=330 v and as the inverter output the voltages

simple Van with amplitude of 220 V.

Figures (2.1) to (2.5) represent the simulations of the inverter output voltages

controlled by different MLI techniques that have been used.

The adjustment coefficient r=0.9 is fixed, and the frequency modulation index mf is varied

for values of 12 and 36.

We note that the simple, symmetric, asymmetric, suboptimal and vector MLIs which

push the harmonic towards the higher ranks. But we note that if we increase the index of

mf modulation by pushing back the strong harmonics more and the THD decreases.

It is noted that the vector and suboptimal PWM have low THD compared to the others.

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
THD = 68.47% THD = 42.08%
150
150
mf=12 mf=36
r=0.9 100 r=0.9
100
Fundamental=70.61% Fundamental=70.75%
Voltage waste = 29.25%
Voltage waste = 29.39%
50 50

0 0
0 10 20 30 (a) 40 50 0 10 20 30 40 50
(b)

Figure (2.1) Simulation of Van voltage by PWM ‡ Natural sampling

• a) Van for r=0.9 and mf=12
• b) Van for r=0.9 and mf=36

24
Machine Translated by Google

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04

THD = 70.26% THD = 42.49%

150 150
mf=12 mf=36
r=0.9 r=0.9
100 100
Fundamental=69.81% Fundamental = 70.55%
Voltage waste = 30.19% Voltage waste = 29.45%
50 50

0 0
0 10 20 30 40 50 0 10 20 30 40 50
(has)
(b)

Figure (2.2) Simulation of the Van voltage by symmetrical PWM

• a) Van for r=0.9 and mf=12
• b) Van for r=0.9 and mf=36

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04

THD = 68.57% THD = 42.45%

150 150
mf=12 mf=36
100 r=0.9 r=0.9
Fundamental=70.48% 100
Fundamental=70.68%
Voltage waste = 29.52% Voltage waste = 29.32%
50 50

0
0
0 10 20 30 40(has)
50 0 10 20 30 40 50
(b)
Figure (2.3) Simulation of Van voltage by asymmetric PWM • a)
Van for r=0.9 and mf=12
• b) Van for r=0.9 and mf=36

25
Machine Translated by Google

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
THD = 69.50% THD = 31.39%

150 150
mf=12 mf=36
100 r=0.9 100 r=0.9
Fundamental=70.48% Fundamental=70.48%
Voltage waste = 29.52% Voltage waste = 29.52%
50 50

0 0
0 10 20 30 40 50 0 10 20 30 40 50
(has) (b)

Figure (2.4) Simulation of Van voltage by suboptimal PWM • a)

Van for r=0.9 and mf=12
• b) Van for r=0.9 and mf=36

500 500

0
0

-500
0 0.01 0.02 0.03 0.04 -500
0 0.01 0.02 0.03 0.04

THD = 88.68% THD = 26.82%

150
150
mf=12 mf=36
r=0.9 r=0.9
100 Fundamental=70.82% 100
Fundamental=70.88%
Voltage waste = 29.18% Voltage waste = 29.12%

50 50

0 0
0 10 20 30 40 50 0 10 20 30 40 50
(b)
(has)

Figure (2.5) Simulation of Van voltage by vector PWM

• a) Van for r=0.9 and mf=12
• b) Van for r=0.9 and mf=36

26
Machine Translated by Google

2.2 Interpretation of curves:

For figures (2.6) up to (2.22) we have represented the simulations of the output voltages
of the inverter. We fix mf=36 and vary the adjustment coefficient.
For simple, symmetric, and asymmetric MLI, we have ar=0.4 and 1. On the other hand, for MLI
vector and suboptimal on ar=0.4 and 1.1547, but the calculated MLI at p=0.4 and 0.9.

We note that increasing the adjustment coefficient to the maximum allows us

to obtain an increase in the fundamental therefore the reduction of voltage waste. We note
also that if the adjustment coefficient increases the THD decreases, and if r decreases the THD will be
important ‡ due to voltage waste.
We note that in the suboptimal and vector PWM we have a greater voltage waste
low compared to other PWMs, we also note that the calculated PWM causes a THD
important.

500 500

0 0

-500
0 0.01 0.02 0.04 -500
0.03
0 0.01 0.02 0.03 0.04
THD = 22.04% THD= 45.01%

150 mf=36
60 mf=36 r=1
r=0.4 100 Fundamental=78.49%
40 Fundamental = 31.16% Voltage waste = 21.51%
Voltage waste = 68.84%
50
20

0
0
0 0 10 20 30 (b) 40 50
10 20 30 40 50
(has)

Figure (2.6) Simulation of Van voltage by PWM ‡ Natural sampling • a) Van for
mf=36 and r=0.4
• b) Van for mf=36 and r=1

27
Machine Translated by Google

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04

THD = 22.29% THD = 44.87%

60 mf=36 150 mf=36

r=0.4 r=1
40 Fundamental = 31.17% 100 Fundamental=78.42%
Voltage waste = 68.83% Voltage waste = 21.58%
20 50

0
0
0 10 20 30 40 50
(has) 0 10 20 30 40 50
(b)
Figure (2.7) Simulation of the Van voltage by symmetrical PWM
• a) Van for mf=36 and r=0.4
• b) Van for mf=36 and r=1

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 THD= 0.04
44.89%
THD = 21.36%

mf=36 150 mf=36

60
r=0.4 r=1
40 Fundamental = 31.67% 100 Fundamental=78.22%
Voltage waste = 68.33% Voltage waste = 21.78%
20 50

0 0
0 10 20 30 40 50 0 10 20 30 40 50
(has) (b)

Figure (2.8) Simulation of the Van voltage by asymmetric PWM •

a) Van for mf=36 and r=0.4
• b) Van for mf=36 and r=1

28
Machine Translated by Google

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
THD = 37.59%
THD = 21.40%
200
60 mf=36 mf=36
r=0.4 r=1.1547
40 Fundamental = 31.57% Fundamental=90.81%
100
Voltage waste = 68.43% Voltage waste = 9.19%
20

0 0
0 10 20 30 40 50 0 10 20 30 40 50
(has) (b)

Figure (2.9) Simulation of Van voltage by suboptimal PWM

• a) Van for mf=36 and r=0.4
• b) Van for mf=36 and r=1.1547

500
500

0
0

-500
0 0.01 0.02 0.03 0.04 -500
0 0.01 0.02 0.03 0.04
,THD= 130.30% THD = 44.69%
200
80 K=4 P=0.4
K=4
Fundamental=40.01%
P=0.9
60 Voltage waste = 59.99%
Fundamental=90%
100 Voltage drop = 10%
40

20

0 0
0 10 20 30 40 50 0 10 20 30 40 50

(has) (b)

Figure (2.10) Simulation of Van voltage by calculated PWM

• a) Van for p=0.4
• b) Van for p=0.9

29
Machine Translated by Google

500 500

0 0

-500 -500
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
THD = 13.22% THD = 32.60%
200
60 mf=36 mf=36
r=0.4 r=1.1547
40 Fundamental = 31.43% Fundamental=90.74%
Voltage waste = 68.57% 100
Voltage waste = 9.26%
20

0 0
0 10 20 30 40 50 0 10 20 30 40 50
(has)
(b)

Figure (2.11) Simulation of Van voltage by vector PWM a) Van

• for mf=36 and r=0.4
• b) Van for mf=36 and r=1.1547

Conclusion

These different simulations revealed to us that:

• the curves given by the suboptimal and vector MLI techniques are closer
better to the real signal.

• as the value of the adjustment coefficient increases, the voltage drop decreases.

• the higher the frequency modulation index, the more the harmonics are pushed back
towards high frequencies forming dharmonic families which are easy to filter by
the sequel.

30
Machine Translated by Google

31